PAPERS
To return to the "Published Papers" main page, click here.
THE EIGHTH INTERNATIONAL CONFERENCE ON FORENSIC COMPUTER SCIENCE - ICoFCS 2013
Print ISBN 978-85-65069-09-9, pages 7-13
DOI: 10.5769/C2013001 and http://dx.doi.org/10.5769/C2013001
A Forensic Tool for Signature Authenticity Verification Through Digital Image Processing and Artificial Neural Networks
By Deivison Pinheiro Franco, Felipe Dantas Barboza and Nágila Magalhães Cardoso
To download this paper, click here.
Print ISBN 978-85-65069-09-9, pages 7-13
DOI: 10.5769/C2013001 and http://dx.doi.org/10.5769/C2013001
A Forensic Tool for Signature Authenticity Verification Through Digital Image Processing and Artificial Neural Networks
By Deivison Pinheiro Franco, Felipe Dantas Barboza and Nágila Magalhães Cardoso
To download this paper, click here.
ABSTRACT
This paper aims to propose a computational forensics tool able to verify the authenticity of handwritten signatures in an automated way, to help and optimize this process and act as a tool for decision support. The methodology of this proposal was based on the use of techniques of digital image processing and neural networks through the backpropagation learning algorithm with 500 and 901 approaches. The results showed an average percentage error of 20% in the first and of 5.83% in the second, and the performance of a trained professional to verify the authenticity of signatures has an average error of 6.67%. Thus, we could observe the efficiency of the proposed tool, as well as the difference and evolution of approaches through the relevance of the results.
KEYWORDS
Authenticity of Signatures, Graphoscopy, Neural Networks, Backpropagation Algorithm, Digital Image Processing.
This paper aims to propose a computational forensics tool able to verify the authenticity of handwritten signatures in an automated way, to help and optimize this process and act as a tool for decision support. The methodology of this proposal was based on the use of techniques of digital image processing and neural networks through the backpropagation learning algorithm with 500 and 901 approaches. The results showed an average percentage error of 20% in the first and of 5.83% in the second, and the performance of a trained professional to verify the authenticity of signatures has an average error of 6.67%. Thus, we could observe the efficiency of the proposed tool, as well as the difference and evolution of approaches through the relevance of the results.
KEYWORDS
Authenticity of Signatures, Graphoscopy, Neural Networks, Backpropagation Algorithm, Digital Image Processing.
